Marginals robo-poll bonanza

Before I dive into today’s glut of electorate-level polling and the picture of unmitigated disaster it paints for Labor, mention should be made of today’s declaration of candidates and determination of ballot paper ordering. I’ve finished labouring through the chore of uploading the candidate lists to my election guide, in the course of which I was unavoidably struck by one salient fact: there are far too many candidates at this election. The total comes in at 1188 for the House of Representatives and 529 for the Senate.

The former number is solidly clear of a previous record of 1109 in 1998, amounting to nearly half an extra candidate per electorate, and well clear of the 849 in 2010, a relatively low number thought to have resulted from the election being called three months ahead of time. The Senate number is still more unprecedented, blowing the lid off the previous record of 367 candidates. Remarkably – suspiciously, even – this comes despite a doubling of nomination deposits to $1000 for House of Representatives candidates and $2000 for Senate candidates.

Some might consider a greater array of candidates a boon for democracy, but in my view that’s entirely negated by the obstacle posed to the act of voting, at least under our present system. This is starkly illustrated by the metre-long Senate ballot papers that voters in the larger states will be required to grapple with on September 7, and the magnifying glasses that will be supplied in polling booths to assist in reading the small print crammed on to them. That will no doubt have all but the tiniest handful of voters opting for the above the line option, exacerbating one of the least attractive features of our system – the mass transfer of votes as dictated by preference deals.

As for the lower house, an analysis by the Australian Electoral Commission indicates that each extra candidate causes a 0.2% increase in the informal vote. If partisan advantage is what matters to you, it’s likely that this makes a large number of candidates disadvantageous to Labor. Labor’s surprise defeat in Greenway at the 2004 election may well have been influenced by an 11.8% informal vote, which was in turn influenced by what I believe to have been a then record (at a general election at least) 14 candidates. This time around there are 12 candidates in Corangamite, Deakin and Mallee, 13 in Bendigo and McMillan, and 16 in Melbourne. Notably, all these electorates are in Victoria, which seems to have the largest number of organised micro-parties – perhaps having been inspired by the example of Family First and the Democratic Labour Party in winning Senate seats over the course of the past decade.

So, to these opinion polls. There are 14 automated phone polls in all from three different agencies, with swings ranging from 0% to 15% and averaging 8%. This is enormously out of kilter with the national polling that was coming through before we hit a dry spell at the start of the week, which suggested a swing of more like 2%. So one might variously hypothesise that there has been a huge shift to the Coalition this week; that the polls have targeted areas where Labor is doing particularly badly; that there may have been something about these polls to bias them towards the Coalition, through some combination of their being automated, mid-week and electorate-level polls; that the national polls have been heavily biased to Labor and the automated polls have shown them up. The latter at least I do not think terribly likely, the truth probably involving some combination of the first three.

We have also had more conventional phone poll results from Newspoll, conducted from Monday to Thursday from samples of 504 each, which oddly target Rob Oakeshott and Tony Windsor’s seats of Lyne and New England. These respectively have the Nationals ahead 59-41 and 66-34, which if anything suggest swings to Labor. The primary votes from Lyne are 26% for Labor, 51% for the Coalition and 7% for the Greens, while from New England it’s 24%, 53% and 5%.

Running through the automated polls:

• Lonergan and JWS Research have both targeted Forde and Lindsay, with very similar results in each case. In Forde, the JWS Research poll of 568 respondents has Liberal National Party member Bert van Manen leading Peter Beattie 54% to 33% on the primary vote and 60-40 on two-party preferred, for a swing of 8.4%. The Lonergan poll, for which The Guardian offers great detail, covered 1160 respondents and showed van Manen’s lead at 56% to 34% and the Greens at just 4%, compared with 12% at the 2010 election. While no two-party preferred figure is provided, it would obviously be very similar to JWS Research’s 60-40. As low as van Manen’s national profile may be, JWS Research gives him a 49% approval rating against 19% disapproval, with Peter Beattie on 35% and 51%. Kevin Rudd’s net approval rating is minus 18% against minus 1% for Tony Abbott. The Lonergan poll has 40% saying Peter Beattie has made them less likely to vote Labor against on 22% for more likely.

• Longergan’s Lindsay poll, conducted on Tuesday night from a sample of 1038, has Liberal candidate Fiona Scott’s primary vote at no less than 60%, up 17% on 2010, with Labor member David Bradbury on 32%, down 13%. The Guardian quotes the pollster saying a question about how respondents voted in 2010 aligned with the actual result – I will assume this took into account the tendency of poll respondents to over-report having voted for the winner. I am a little more puzzled by the claimed margin of error of 3.7%, which should be more like 3% given the published sample size (UPDATE: It transpires that this is because Lonergan has, unusually, done the right thing – calculate an effective margin of error that accounts for the fact that the sample is weighted, and that cohorts within it have been extrapolated from sub-par samples). The JWS Research result has the primary votes at 57% for Liberal and 35% for Labor, with two-party preferred at 60.7-39.3.

• ReachTEL has four polls with samples of around 600 apiece, which have the Liberals leading 65-35 in Bennelong (a swing of about 12%) and 53-47 in McMahon (11%) and 52-48 in Kingsford Smith (7%), with Labor hanging on by 52-48 in Blaxland (10%).

• The other Financial Review/JWS Research results show the Coalition ahead in Brisbane (54.1-45.9 from primaries of 50% LNP, 36% Labor), Macquarie (55.1-44.9, 51% Liberal, 35% Labor), Corangamite (53.3-46.7, 48% Liberal, 36% Labor), Aston (63.4-36.6, Liberal 59%, Labor 29%), and Banks (52.8-47.2, Liberal 50%, Labor 43%). The one ray of sunlight for Labor is their 51-49 lead in Greenway, from primaries of 46% for Liberal and 44% for Labor. A full graphic of the JWS Research results is available from GhostWhoVotes, including some diverting results on personal approval. Bert van Manen in Forde and Alan Tudge in Aston appear to rate as very popular local members, while David Bradbury in Lindsay and Darren Cheeseman in Corangamite do not. And Fiona Scott in Lindsay, fresh from the publicity bestowed upon her by Tony Abbott, is easily the highest rating of the challengers.

Author: William Bowe

William Bowe is a Perth-based election analyst and occasional teacher of political science. His blog, The Poll Bludger, has existed in one form or another since 2004, and is one of the most heavily trafficked websites on Australian politics. View all posts by William Bowe